Model-Based Reinforcement Learning Exploiting State-Action Equivalence.
Mahsa AsadiMohammad Sadegh TalebiHippolyte BourelOdalric-Ambrym MaillardPublished in: ACML (2019)
Keyphrases
- model based reinforcement learning
- state action
- reinforcement learning
- markov decision processes
- average reward
- action space
- markov decision process
- stochastic games
- evaluation function
- reward function
- optimal policy
- state transitions
- state space
- function approximation
- learning algorithm
- infinite horizon
- knowledge base
- partially observable
- reinforcement learning algorithms
- function approximators
- machine learning
- partially observable markov decision processes
- temporal difference
- finite state
- multi agent